Ultra low noise amplifier for coil sensor emitting very small signal

[ricko_uk]

Hi,
this is just something I am playing with as an experiment of an idea I had... I'm trying to use a simple air coil to detect an extremely weakly magnetized sphere moving. The signal amplitude is so tiny that even with an oscilloscope you can barely see that something is happening in the noise floor but cannot really distinguish the signal itself, what kind of off-the-shelf amplifier would you use?

Jfet or instrumentation amplifier dev kit? Or some other solution?

If a JFET or INA, which in-amp would you suggest (that perhaps is designed for reading coils)?

Thank you

[TimFox]

What is the resistance of your coil, and your estimate of the tiny voltage from your noisy 'scope trace?
Assuming a very low resistance (say, < 100 ) you probably want a good BJT rather than a JFET (which are better for high source impedance).
Have you looked at the preamplifiers sold by THAT corp. for use with microphones?   https://thatcorp.com/low-noise-preamplifier-ics/
Also, you need to seriously reduce the bandwidth of your system to reduce the noise:  how low can you go for your signal application?

[shabaz]

Just to throw another idea out there; if you don't mind that moving metals (magnetized or not) will be detected as well, then TI's LDC1000 is very good.
Photo below shows the PCB coil being used to detect the movement of the clock's second hand as it sweeps past every minute.

[Sensorcat]

Just to throw another idea out there; if you don't mind that moving metals (magnetized or not) will be detected as well, then TI's LDC1000 is very good.

That's a very good and easy to use solution if the item to detect moves slowly, i.e. with low frequency. Due to the resonant nature of the operating principle of the LDC1000 (and its siblings), it cannot track an item moving with hundreds of Hz.

[T3sl4co1l]

Well, that's very promising news then: oscilloscope noise floors are notoriously awful, having 100s of MHz bandwidth and typical 8 bit acquisition.  If you can measure a few mV, you're doing well!  That means only a few hundred or thousand gain will be required here, and the noise figure at low frequencies can be quite good using a chopper or auto-zero amp.  Do make sure the coil is well filtered, so it isn't picking up EMI that saturates the amp.

JFET or CMOS really isn't the way to go here, as the impedance near DC is low (coil DCR), you're presumably well below the L/R time constant so it looks like a Thevenin equivalent source.  Those parts usually have a higher 1/f knee frequency as well.  You could use a BJT amplifier for sub-1nV/rtHz noise, but compensating thermal offsets is nontrivial (short of a monolithic pair).  CMOS autozero amps really do quite excellent these days, and in any case the noise floor will be far better than the scope's (uV vs. mV) so the SNR should be just fine.

Tim

[PCB.Wiz]

Hi,
this is just something I am playing with as an experiment of an idea I had... I'm trying to use a simple air coil to detect an extremely weakly magnetized sphere moving. The signal amplitude is so tiny that even with an oscilloscope you can barely see that something is happening in the noise floor but cannot really distinguish the signal itself, what kind of off-the-shelf amplifier would you use?

How fast/far is this moving, over what range of numbers ?
Magnetic induction is speed proportional, so very slow movements will mean very low signals.
As indicated above capacitive, or differential capacitive, may be better, as that works even on a static metal object.

[Sensorcat]

Magnetic induction is speed proportional, so very slow movements will mean very low signals.

Only if the coil system is used as a generator. If it is powered and the impedance measured, there is no speed proportionality. The LDC1000 proposed by shabaz does that by measuring the frequency of an LC oscillator with the sensor coil.

[Dr. Frank]

Hi,
this is just something I am playing with as an experiment of an idea I had... I'm trying to use a simple air coil to detect an extremely weakly magnetized sphere moving. The signal amplitude is so tiny that even with an oscilloscope you can barely see that something is happening in the noise floor but cannot really distinguish the signal itself, what kind of off-the-shelf amplifier would you use?

Jfet or instrumentation amplifier dev kit? Or some other solution?

If a JFET or INA, which in-amp would you suggest (that perhaps is designed for reading coils)?

Thank you

As you already stated, the signal is barely visible in the noise.
Therefore, an amplifier will not do the job at all, because it amplifies the noise in the same manner.

I suggest to 1st set up your pick-up coil in a bridge configuration, and 2nd to use a Phase-Locked method to detect the movement of your magnet.
Maybe it's possible to move the magnet in a repetitive manner, then you could synchronize to the frequency of your moving system.

I made magnetic AC measurements on superconductors, as well on very small samples (100µm x 300µm), using a so called Hartshorn bridge.
The differential bridge output AC signal was in the mV, down to the nV range.
So I used 100x differential low noise amplifiers first, to achieve an SNR of up to 108dB @ mV input signal level, for digitization with my 3458A.
The LI was of course capable to achieve that SNR on even smaller signals.

I also designed one low-noise differential amplifier, but that's over 30 years ago .. I'll have to look, which IC  I have used, and if there is a successor available.   

It was an SSM2016, from PMI, 1988 vintage.
Low noise, low distortion amplifier.

Frank

[David Hess]

A bipolar low noise precision operational amplifier would be my first choice, like an OP-27/LT1007, or even AD797/LT1128, but as T3sl4co1l points out, a modern chopper stabilized part would also work well.

There are modern JFET operational amplifiers with low voltage noise which could also be used, but the bipolar parts are still lower noise.

[ricko_uk]

Thank you all, that gives me quite a few things to look into, try and compare the performance of.

When trying the solutions using amplifiers, do I just connect the coil straight across the amplifiers inputs or do I need some passives in front of it, either for filtering and/or resistors to ground to provide DC return path like shown on page 55 in this Instrumentation Amplifier document (I know we are not using an INA but was wondering if the same to the other suggested types of amplifiers)? https://www.analog.com/media/en/training-seminars/design-handbooks/designers-guide-instrument-amps-complete.pdf

Or perhaps a resistor in parallel to the coil like I have seen in some schematics like this one? https://www.circuitlab.com/circuit/4x8brgg55j73/ae-week-9-op-amp-386-coil/

Thank you

[David Hess]

When trying the solutions using amplifiers, do I just connect the coil straight across the amplifiers inputs or do I need some passives in front of it, either for filtering and/or resistors to ground to provide DC return path like shown on page 55 in this Instrumentation Amplifier document (I know we are not using an INA but was wondering if the same to the other suggested types of amplifiers)? https://www.analog.com/media/en/training-seminars/design-handbooks/designers-guide-instrument-amps-complete.pdf

Each input requires a DC path to absorb input bias current.  The coil will do this just fine as long as one side is grounded, as shown below.  If the coil is floating and used with an instrumentation amplifier, then a resistor from one side of the coil to ground or some other potential within the common mode range of the instrumentation amplifier would be required.

RIAA magnetic coil amplifiers, as shown below, provide an example of what you could do, but without the RIAA equalization network and AC coupling, which is shown in the second example.  The coil is directly connected to a non-inverting amplifier, and the 2 resistor feedback network controls the gain.  As shown the gain is 11 but the 1 kilohm resistor can be increased to increase the gain.

Amplifier selection for lowest noise depends on DC resistance of the coil.  In general coil resistance is low leading to parts like the LT1028 being the best choice.  If your coil has hundreds of ohms of resistance, then an operational amplifier like the LT1007 or OP-27 would be better.

Or perhaps a resistor in parallel to the coil like I have seen in some schematics like this one? https://www.circuitlab.com/circuit/4x8brgg55j73/ae-week-9-op-amp-386-coil/

That is a poteniometer being used as a variable attenuator for gain control.  You can do that, but it will add noise.  Better would be to use the poteniometer on the output of the first stage of amplification if this is acceptable.